/AlphaTensor /wikipedia |
https://twitter.com/espadrine/status/1670128880328343552?s=20
1. whois
But AMD now has the same trick. In RDNA3, with WMMA, the RX 7900 XTX has instructions that do a 16×16 matrix multiplication in 16 cycles in a single Compute Unit (64 threads). 2×16^3 (matmul ops) ÷ 16 (cycles) = 512 ops/CU/cycle.
https://twitter.com/JosephJacks_/status/1669750155518808064?s=20
2. history
Here’s the latest OpenAI regulatory capture propaganda pushing the government to make it illegal for anyone to do advanced matrix multiplication aka use computers…
Computational complexity of matrix multiplication https://en.wikipedia.org/wiki/Computational_complexity_of_matrix_multiplication